rapid dopaminergic and gabaergic modulation of calcium and voltage transients in dendrites of...

J Physiol 00.00 (2012) pp 1–21 1

The

Jou

rnal

of

Phys

iolo

gy

Rapid dopaminergic and GABAergic modulation of calciumand voltage transients in dendrites of prefrontal cortexpyramidal neurons

Wen-Liang Zhou and Srdjan D. Antic

Department of Neuroscience, University of Connecticut Health Center, 263 Farmington Avenue, Farmington, CT 06030, USA

Key points

• Dopamine-releasing axons make direct synaptic contacts with the dendrites of corticalpyramidal neurons. It is not known if and how dendritic physiology changes upon dopaminerelease at these synapses.

• We attempted to mimic synaptically released dopamine by ejecting dopamine from a micro-pipette. Action potential-induced voltage transients and corresponding calcium influx wereboth measured in thin dendritic branches using voltage-sensitive and calcium-sensitive dyes,before and after local application of dopamine or GABA.

• GABA blocked calcium influx in dendrites by blocking AP backpropagation. Dopamine, onthe other hand, reduced dendritic calcium influx only at the site of dopamine release. APssuccessfully propagated through the dopamine application site.

• Dopamine blocked dendritic voltage-gated calcium channels in the presence of protein kinaseblockers, suggesting a membrane delimited mechanism.

• Spatially restricted dopamine-mediated suppression of dendritic calcium is expected to occurduring phasic dopaminergic signalling, when midbrain dopaminergic neurons respond to asalient event.

Abstract The physiological responses of dendrites to dopaminergic inputs are poorly under-stood and controversial. We applied dopamine on one dendritic branch while simultaneouslymonitoring action potentials (APs) from multiple dendrites using either calcium-sensitive dye,voltage-sensitive dye or both. Dopaminergic suppression of dendritic calcium transients was rapid(<0.5 s) and restricted to the site of dopamine application. Voltage waveforms of backpropagatingAPs were minimally altered in the same dendrites where dopamine was confirmed to cause largesuppression of calcium signals, as determined by dual voltage and calcium imaging. The dopamineeffects on dendritic calcium transients were fully mimicked by D1 agonists, partially reduced byD1 antagonist and completely insensitive to protein kinase blockade; consistent with a membranedelimited mechanism. This dopamine effect was unaltered in the presence of L-, R- and T-typecalcium channel blockers. The somatic excitability (i.e. AP firing) was not affected by strongdopaminergic stimulation of dendrites. Dopamine and GABA were then sequentially applied onthe same dendrite. In contrast to dopamine, the pulses of GABA prohibited AP backpropagationdistally from the application site, even in neurons with natural Cl− concentration (patch pipetteremoved). Thus, the neocortex employs at least two distinct mechanisms (dopamine and GABA)for rapid modulation of dendritic calcium influx. The spatio-temporal pattern of dendriticcalcium suppression described in this paper is expected to occur during phasic dopaminergic

C© 2012 The Authors. The Journal of Physiology C© 2012 The Physiological Society DOI: 10.1113/jphysiol.2011.227157

2 W.-L. Zhou and S. D. Antic J Physiol 00.00

signalling, when midbrain dopaminergic neurons generate a transient (0.5 s) burst of APs inresponse to a salient event.

(Resubmitted 25 December 2011; accepted after revision 25 May 2012; first published online 28 May 2012)Corresponding author S. D. Antic: UConn Health Center, Neuroscience, Rm E-3038, 263 Farmington Avenue,Farmington, CT 06030-3401, USA. Email: [email protected]

Abbreviations ACSF, artificial cerebrospinal fluid; AP, action potential; DA, dopamine; IPC, intrapipetteconcentration, as defined in the text; PFC, prefrontal cortex; ROI, region of interest, as defined in the text; VGCC,voltage-gated calcium channel.

Introduction

More than 20 years after the discovery that dopaminergicsynapses populate the dendrites of pyramidal neurons inthe prefrontal cortex (PFC) (Goldman-Rakic et al. 1989),it is still unclear what physiological changes occur indendrites upon a transient (synaptic-like) dopaminergicstimulus (Schultz, 2002; Seamans & Yang, 2004). In vitroexperiments investigating changes in dendritic AP signalsin the PFC relied on bath applications of dopamine (DA)and DAergic drugs (Gulledge & Stuart, 2003; Kisilevskyet al. 2008). There is an obvious temporal inconsistencybetween naturally occurring phasic DAergic transients(Ljungberg et al. 1992) and bath application of DA.In a typical bath application experiment, the change(increase or decrease) in DA concentration is very slow,persists for many minutes, and non-selectively activatesall DA-sensitive elements in the brain tissue, includingdendrites, axons, somata of various neuron types, andeven non-neuronal elements such as glia and bloodvessels. In summary, with bath application of DA, twocritical features of natural DAergic neurotransmissionare lost: (a) rapid and transient character (Ljungberget al. 1992; Schultz, 2002) and (b) the spatial selectivity(Goldman-Rakic et al. 1989).

Using a bath-application approach, Gulledge andStuart (2003) reported that DA did not modulate back-propagating APs nor AP-induced Ca2+ transients. This wasa surprising result as DA modulates both voltage-gatedsodium channels (Maurice et al. 2001; Peterson et al.2006; Few et al. 2007) and voltage-gated calcium channels(Malgaroli et al. 1987; Lledo et al. 1990; Surmeier et al.1995; Young & Yang, 2004; Bender et al. 2010). In acounterpart study, also conducted in rat PFC, a globaland prolonged application of DA (5 min) significantlyreduced AP-induced calcium transients in the apicaltrunk (Kisilevsky et al. 2008). However, several importantquestions remain unanswered. (1) What is the spatialprofile of the DA effect? (2) What time interval passesbetween DA release and the actual change in dendriticCa2+? (3) Are DA-induced changes in AP-associateddendritic Ca2+ transients caused by alterations in thedendritic AP voltage waveform? (4) How does DAergicinhibition compare to GABAergic inhibition on the same

cell? (5) Which subclass of pyramidal dendrites (apical orbasal) is more sensitive to DAergic modulation? Previousresearch on DAergic modulation of dendritic functionhas focused exclusively on the apical dendrite (Yang &Seamans, 1996; Hoffman & Johnston, 1999; Gulledge &Stuart, 2003; Kisilevsky et al. 2008). Since, in the ratPFC, the densities of DAergic fibres and DA receptors aregreatest in deep layers V and VI (Sesack et al. 1989), wherebasal dendrites reside, it is possible that basal dendritesmay be more susceptible to DA modulation than apicaldendrites.

Rapid and transient increases in DA concentrationare expected to occur during phasic DAergic signalling,when midbrain DAergic neurons generate a transient(0.5 s) burst of action potentials in response to asalient event in the animal’s environment (Ljungberget al. 1992; Schultz, 2002). Here we used fast multi-sitecalcium and voltage imaging to evaluate how a briefincrease in DA concentration, presented locally withinone spatially restricted segment of a dendritic branch,affects both calcium and voltage dynamics in apical andbasilar dendrites of the rat PFC pyramidal neurons.Dendritic calcium and voltage signals were recordedoptically using intracellularly applied calcium-sensitive orvoltage-sensitive dyes, either individually or concurrently.By monitoring dendritic function simultaneously frommany dendritic segments, our experiments addressed thespatial extent of DAergic inhibition of dendritic Ca2+

influx along the target branch, as well as in neighbouringdendritic branches. Transient pressure applications ofDA were used to determine the latency between theDA stimulus onset and the ensuing change in dendriticcalcium amplitude. DAergic modulations of calcium orvoltage signals were directly compared to GABAergicmodulations by sequentially applying DA and thenGABA (or vice versa) on the same dendrite of the samepyramidal neuron. To preserve the natural concentrationof chloride ions, patch electrodes were removed beforeGABA applications. The results of these experimentssuggested that the mammalian PFC employs at leasttwo distinct forms of fast modulation of dendritic Ca2+,based on two distinct neurotransmitter systems, DA andGABA.

C© 2012 The Authors. The Journal of Physiology C© 2012 The Physiological Society

J Physiol 00.00 Dopaminergic and GABAergic modulation in dendrites 3

Methods

Brain slice and electrophysiology

Sprague–Dawley rats (P21–42) were anaesthetized withisoflurane inhalation, decapitated and the brains extractedwith the head immersed in ice-cold, artificial cerebrospinalfluid (ACSF), according to an animal protocol approvedby the Center for Laboratory Animal Care, Universityof Connecticut. ACSF contained (in mM): 125 NaCl, 26NaHCO3, 10 glucose, 2.3 KCl, 1.26 KH2PO4, 2 CaCl2

and 1 MgSO4, pH 7.4. Coronal slices (300 μm) werecut from frontal lobes. All experimental measurementswere performed at 32–34◦C. Whole-cell recordingswere made from visually identified layer V pyramidalneurons within the ventral medial PFC, including pre-limbic and infralimbic cortex. Intracellular solutioncontained (in mM): 135 potassium gluconate, 2 MgCl2,3 Na2-ATP, 10 Na2-phosphocreatine, 0.3 Na2-GTP and 10Hepes (pH 7.3). Electrical signals were amplified with aMulticlamp 700B and digitised with two input boards:(1) Digidata Series 1322A (Molecular Devices, UnionCity, CA, USA) and (2) Neuroplex (RedShirtImaging,Decatur, GA, USA). Only cells with a membrane potentialmore negative than −50 mV (not corrected for junctionpotential) and action potential amplitudes exceeding80 mV (measured from the baseline) were included in thisstudy. Action potentials were evoked with depolarizingcurrent steps injected into the cell body (intensity1.0–1.5 nA; duration, 1.5–2.0 ms). To generate trains ofsomatic APs at ∼20 Hz, a depolarizing current pulse(intensity, 100–200 pA; duration, 800 ms) was injectedinto the soma. Dendritic recordings were performed usinghigh-resistance patch electrodes (∼15 M�). To generateAPs in unpatched neurons stimulating electrodes wereplaced close to the axon.

Dye injection

Voltage-sensitive dye (JPW-3028) and calcium-sensitivedyes (Ca-Green-1, Oregon Green BAPTA-1 and bis-fura-2;Invitrogen, Carlsbad, CA, USA) were dissolved in intra-cellular solution. The protocol for voltage-sensitive dyeinjection has been previously described (Antic, 2003).Briefly, neurons were filled through whole-cell patchingpipettes with JPW-3028 for 45 min. Dye-free solutionwas at the tip of the pipette, while the back ofthe pipette lumen was loaded with dye-rich solution(400–800 μM). The filling pipette was carefully pulledout (outside-out patch) and brain slices were left toincubate for 1–3 h at room temperature. Immediatelybefore optical recordings the cells were re-patched witha dye-free pipette. In calcium imaging experiments, AlexaFluor 594 (50 μM) was included in the intracellularsolution to aid the visually guided focal application

of drugs onto dendrites. The calcium-sensitive dyes(100–150 μM) were injected for 30–45 min before opticalrecordings.

Dendritic voltage and calcium imaging

Multi-site dendritic imaging was performed on eitheran Olympus BX51WI or Zeiss, Axioskop FS micro-scope, equipped with UV-compatible 40× objective, twocamera ports and a low-ripple xenon arc lamp (OptiQuip,Highland Mills, NY, USA) for epi-illumination.Functional dendritic imaging was performed with aNeuroCCD camera (80 × 80 pixels, RedShirtImaging).Voltage-sensitive dye signals were sampled at a 2000or 2700 Hz frame rate, and calcium-dye signals at500 Hz. In voltage–calcium dual-imaging experiments,the signals were first recorded (control, drug, washout)with one filter set (cube), and then experimenterswould change the filter cube and repeat the entireprotocol (control, drug, washout). Between recordings,all stimulus parameters (location, intensity and duration)were kept the same. A spike-triggered averaging routineavailable in Neuroplex (RedShirtImaging) was used toimprove the signal-to-noise ratio of voltage-sensitive dyemeasurements. The same number of sweeps was used incontrol, drug and wash. The number of individual sweepsused for averaging did not exceed four. One DA puff wasapplied prior to each individual sweep included in theaveraging procedure.

Drug application

Dopaminergic drugs were dissolved in ACSF and loadedinto glass micropipettes (6–8 M�). Tips of the drugapplication pipettes were positioned ∼20 μm fromthe dendritic shaft. Drugs were pressure-ejected viaa computer-driven picospritzer (puff duration, 3 s)prior to a calcium imaging measurement, or for20–500 ms (duration) during the optical measurement.For iontophoretic application, dopamine (20–100 mM,pH 4.0) was loaded into a sharp micropipette (∼40 M�)and driven out by positive current (nominal intensity,1.0–1.8 μA; duration, 20–300 ms). All drugs (SKF-38393(SKF), A-68930 (A6), SCH23390 (SCH), quinpirole(QP), GABA) were purchased from Sigma (St Louis,MO, USA). The intrapipette concentration (IPC)refers to the concentration of drug inside a glassmicropipette.

Data analysis

Analysis of optical data, including spatial averaging,high-pass and low-pass filtering, was conducted withNeuroplex 8.0.0 (RedShirtImaging). To process off-linecalcium imaging data, we applied a Butterworth high-passfilter at 0.1 Hz cut-off and a Gaussian low-pass filter



at 90 Hz cut-off; for voltage imaging data, the high-and low-pass filters were set at Butterworth 0.1 Hz andGaussian 900 Hz, unless otherwise specified. Pressureejection of drugs during optical imaging causes a trans-ient reduction in resting light intensity (RLI), whichwe attribute to a recording artifact. The artifact sub-traction procedure is described in Supplemental Fig. S3.In experiments where DA was applied prior to the openingof the shutter, raw traces (no subtraction) were usedin the analysis. Electrical recordings were analysed inClampfit 9 (Molecular Devices, Sunnyvale, CA, USA). Foroptical recordings, a custom-made MATLAB routine wasused to automatically calculate AP half-widths and APintegral (area under the curve). This algorithm included:spatial averaging of pixel outputs in a pre-selected ROI,low-pass filtering of the optical signal (980 Hz), inter-polation of the optical signal (Fig. 3G in Antic, 2003),finding of the AP peak, and calculation of the baselinelevel by averaging data points prior to the AP onset.Statistical tests were performed using SigmaPlot 8.02(Systat Software, San Jose, CA, USA). All statistics weredone on raw data points before normalization unlessotherwise specified. Paired Student’s t tests were used forcomparing data obtained from the same neuron (in twodifferent conditions). Unpaired Student’s t tests were usedfor data obtained from different neurons. Significancewas set as ∗P < 0.05, and high significance as ∗∗P < 0.01.Values are presented as mean ± SEM. The exclusion ofnon-responding cells from the data set did not affectour conclusions because non-responders comprised arelatively small fraction of the total (9 out of 64 pyramidalneurons). For example, 3 out of 16 cells were resistant to10 mM DA pulse (signal reduction <10%). In both datagroup versions n = 16 (with non-responders) and n = 13(without non-responders) the difference between DA andControl was statistically significant (P < 0.01). Fittingof the dopamine dose–response curve (SupplementalFig. S1) was done in SigmaPlot using the equation:f = min + (max − min)/(1 + (x / EC50)Hillslope), where xis intrapipette concentration of DA, min is the reductionin the calcium signal at 1 μM, and max is the reduction inthe calcium signal at 20,000 μM. To calculate the ‘spaceconstant’ for decay of DA effect along dendrite, theDA/Control ratios were (1) normalized in respect to thevalues obtained at the DA application sites, (2) inverted,and then (3) fitted in SigmaPlot using the followingequation:

E x = E 0 × e(− xλ

),

where E0 is the amplitude of DA effect at the DAapplication site; Ex is amplitude of DA effect at distance‘x’; λ is ‘length or space constant’ defined as a distancefrom the DA application site at which DA effect falls downto 37% of its initial value at x = 0.

Results

Local and brief dopamine application

Brief puffs of DA (duration, 3 s) were applied locallyto basal dendrites. DA puffs were aimed at the middlesegments of basal dendrites (Fig. 1AB, ROI 2), at anaverage distance of 156 ± 40 μm from the soma (N = 65basal dendrites; n = 35 neurons). Optical recordings ofAP-induced dendritic Ca2+ transients were first acquiredjust before drug application (Fig. 1C, Control), and then2 s after closing of the pressure valve (Fig. 1C, Dopamine),using the same pattern of somatic APs (ROI 0). In responseto DA (1 mM inside micropipette), the amplitude of thefirst AP-induced dendritic signal in the train (Fig. 1C,Dopamine, arrow) was on average 67.9 ± 7.1% (n = 7neurons, N = 16 dendrites; P < 0.01, paired t test) of itsoriginal size obtained prior to the DA puff (Control). TheDA effect was short lived, taking less than 90 s for completerecovery of the DA effect in all dendritic loci (Fig. 1C,Washout).

We systematically varied the concentration of dopamineinside the application pipette from 1 μM to 20,000 μM:1 μM (n = 3 cells, N = 5 dendrites), 10 μM (n = 3, N = 4),100 μM (n = 4, N = 11), 1000 μM (n = 7, N = 16),10 000 μM (n = 13, N = 17) and 20,000 μM (n = 5,N = 8). The effect of DA increased with an increase inthe intrapipette concentration (IPC) of DA (Fig. 1D). TheDA dose–response curve (Fig. 1D) was fitted as describedin the Methods, and the EC50 value (325.7 ± 98 μM)was determined from the best-fit curve (SupplementalFig. S1D3).

Local DA applications caused changes in fluorescenceintensity before the arrival of an AP (resting light level, F),suggesting that DA affects Ca2+ under basal conditions(Malgaroli et al. 1987; Akerman et al. 1991). However,neither the direction of change (decrease or increase in F),nor the amplitude of change in F (Supplemental Fig. 1D2)could account for the DA-induced changes in Ca2+ signalamplitude, expressed as �F/F (Supplemental Fig. 1D1).Also, optical measurements performed in three pyramidalcells (6 dendritic branches) treated with a puff of ACSF(duration 3 s) failed to affect the amplitudes of AP-inducedoptical signals (103 ± 6%, P > 0.05, paired t test). Thesedata indicate that local and transient (3 s) surges in DAconcentration reduce AP-evoked dendritic Ca2+ influx.

Non-responders

In 9 out of 64 pyramidal neurons tested, the applicationsof relatively high concentrations of DA (e.g. 1–20 mM

IPC) did not induce any significant changes onAP-induced dendritic Ca2+ signals. The occurrenceof dopamine-insensitive neurons in the mammalianneocortex and neostriatum is common, but is stillan unexplained phenomenon (Sesack & Bunney, 1989;



Surmeier et al. 1995; Day et al. 2008; Kisilevsky et al. 2008;Moore et al. 2011). In the present study DA-insensitiveneurons (n = 9) were excluded from quantifications.

Spatial aspect of dopaminergic modulation ofdendritic Ca2+

In four neurons, the geometry of the target basaldendrite allowed us to simultaneously monitor ∼80 μmof dendritic length on both sides of the DA applicationsite (Fig. 2A, drawing; Supplemental Fig. S2). When DApulses were applied to the mid-dendrite region (Fig. 2A,0 μm), the dendritic segments located proximally, as wellas distally to the DA application site, experienced little orno change in Ca2+ signal amplitude (Fig. 2A, DA/Ctrl).The DA effect was maximal at the DA application site(Fig. 2A, 0 μm). At a distance of 31.8 ± 2.9 μm away fromthe DA application site, the amplitude of the DA effectdecayed to 37% of its value at the DA application site(n = 4). The effects of DA outside of DA applicationsite were instantaneous and thus unlikely to representdiffusion of an intracellular messenger along the dendriticaxis. In 35 neurons, the DA effect was not observed inthe neighbouring dendritic branches (Fig. 1C, ROI 3), noteven after the washout period (∼90 s), suggesting that localDA stimulation does not generate a diffusible intracellularmessenger that can travel between two neighbouringbranches (Supplemental Fig. S1, ROIs 5 and 6), or tworemote segments of the same dendritic branch (Fig. 2A).

Temporal aspect of dopaminergic modulation ofdendritic Ca2+

In order to detect the earliest DA-induced change in APsignal, in the next series of experiments we applied DA

during a train of APs (Fig. 2B, Supplemental Fig. S3).Within the first 500 ms after the onset of a DA puff (DAcommand pulse), the AP-associated dendritic Ca2+ trans-ients were notably suppressed (Fig. 2B, time point markedby a filled circle, ∼0.5 s). The quantification of data wasperformed on three time points (three APs) marked byfilled circles, triangles and diamonds in Fig. 2B (DuringDA, Short Delay and Longer Delay, respectively). Duringactive ejection of DA (During DA), the AP-associatedCa2+ signals were significantly smaller than control valuesin the same cell (Fig. 2B, During DA, grey column),as determined by the paired t test on raw data points(�F/F), before normalization (P < 0.01, n = 10). Calciumsignals were also significantly smaller than control signalsshortly after the cessation of the DA pulse, filled triangle,Short Delay, ∼2 s (P < 0.01), but were not different fromcontrol values (dotted line 100%) at a longer time delay(∼7 s) following DA application (filled diamond, LongerDelay, P > 0.05, n = 10). These data showed that the onsetof the DA effect on the dendritic Ca2+ transient wasrapid (<500 ms) and short-lived. The DA effect was notdetectable 7 s after the cessation of the DA pulse (LongerDelay). In other words, the effect of DA on dendritic Ca2+

transients was manifested only during the time periodwhen DA molecules were available in the vicinity of thedendrite (during DA pulse and immediately after thecessation of DA pulse). This result also suggests that theobserved spatial profile of the DA effect (Fig. 2A, graph)represents a concentration gradient of DA ejected fromthe DA application pipette.

When drug-free ACSF was ejected onto the basaldendrite, the amplitudes of AP-mediated Ca2+ trans-ients were not significantly different from controlmeasurements at any of the three chosen time points (filled

Figure 1. Local DA application suppresses AP-mediated Ca2+ transientsA, microphotograph of a pyramidal neuron filled with OGB-1. B, one movie frame captured by fast CCD camera(500 Hz, 80 × 80 pixel). Drawing marks the position of DA-filled pipette (1 mM) inserted into the brain slice.C, somatic current injections were used to produce 3 action potentials. Electrical signal from the soma (regionof interest, ROI, 0) is aligned with dendritic recordings from three ROIs (1–3) shown in B. Upon application ofdopamine (DA puff) the amplitude of the first AP-mediated Ca2+ transient at ROI 2 (Dopamine) is less than thatin the measurement obtained just before DA puff (Control) or 90 s after DA puff (Washout). Arrow marks theoptical trace showing notable reduction in Ca2+ signal compared with Control. D, Ca2+ signal decreases with anincrease in the concentration of DA inside the application pipette, shown on a logarithmic scale.



circles, triangles and diamonds; Fig. 2B, white columns),thus indicating that mechanical artifacts alone could notaccount for the observed changes in signal size.

Dopamine receptor agonists

When applied locally for 3 s (Fig. 3A), the D1 receptoragonist SKF-38393, at lower concentrations (IPC, 50 μM),did not have an effect on Ca2+ transients (88.7 ± 7.4%,n = 3; N = 3; P > 0.05, paired t test). However, at anIPC of 500 μM, SKF markedly reduced AP-dependentCa2+ influx in the target segments of basal branches to60.8 ± 5.2% of its original size (n = 5, N = 9, P < 0.01,paired t test; Fig. 3B and C). Similar results wereobtained with another D1 receptor agonist, A-68930 (IPC,500 μM; 66.7 ± 3.5%; n = 3; N = 6, P < 0.01, paired t test;Fig. 3C). While both D1 agonists SKF-38393 and A-68930(IPC, 500 μM) robustly and reliably suppressed dendriticCa2+ transients, the D2 receptor agonist quinpirole (IPC,500 μM) failed to induce any changes (104.7 ± 11.9%,n = 4; N = 7; P > 0.05, paired t test). Quinpirole had

no effect on dendritic Ca2+ transients even at an IPC of5000 μM (102.0 ± 6.1%, n = 3; N = 7; P > 0.05, pairedt test; Fig. 3D). These data indicate that stimulation ofD1-like, but not D2-like family receptors, leads to thesuppression of AP-evoked dendritic Ca2+ influx in basilardendrites of PFC pyramidal neurons. An alternative inter-pretation of these data is that drugs with an affinity forD1-like receptors (dopamine, SKF-38393 and A-68930)directly impede the function of dendritic voltage-gatedcalcium channels, when applied at high concentration (e.g.>325 μM, based on DA EC50).

D1 dopamine receptor antagonist

In the next series of experiments we tested whetherthe D1 receptor antagonist SCH23390 (SCH) can blockthe effect of DA. To avoid a possible contamination ofdata by non-responder cells, we first verified cells wereresponsive to DA in regular ACSF (Fig. 3E, DA in ACSF;Supplemental Fig. S4B1). Only if the cell responded toDA, SCH (150 μM) was introduced in the bath and

Figure 2. Spatial and temporal dynamics of thedopamine effectA, drawing depicts a DA-filled pipette near one basalbranch. Dashed boxes indicate regions of interest alongthe dendrite. AP-induced Ca2+ transients were recordedsimultaneously at multiple ROIs before (Control) and500 ms after a brief (100 ms duration) iontophoreticejection of DA (IPC, 20 mM). Quantification of data wasperformed by calculating the amplitude ratio betweenDA and Control for each region of interest (ROI) andaveraged across 4 neurons. X-axis indicates distancebetween ROI and DA application site. Horizontal errorbars describe variation in ROI location between 4neurons. Vertical error bars describe variation in DA/Ctrlratio between 4 neurons at corresponding ROIs. B, sameexperimental outline as in A, except DA IPC was 10 mM,and optical recordings are from ROI at the DAapplication site. In the control sweep (Ctrl, dashed greyline) the cell body was injected with short current pulsesto produce a train of 23 somatic APs (2.5 Hz). In thenext recording APs were paired with dopamine pressurepulse (DA, thick black line). The command signal fordriving the picospritzer valve shows the exact timing ofthe DA pulse. Dendritic Ca2+ transients aresuperimposed to show the difference between control(grey) and DA sweep (black). Grey columns: Ca2+ signalamplitudes were quantified at three characteristic timepoints marked on optical traces by •, � and ♦. Thesignal amplitude upon DA treatment was normalized tothe control measurement (before DA treatment) andaveraged (n = 10 neurons). Asterisks compareDA-effects against the matching controls (dashed line100%) using a paired t test on raw data (beforenormalization). Open columns: the same as in greycolumns except experiments were repeated by omittingDA from the application pipette (ACSF, n = 3).



experimental measurements were repeated on the samecell (Fig. 3E, DA in SCH). The amplitudes of dendriticAP-associated Ca2+ transients were significantly reducedby DA, both in the absence (P < 0.01) and in the presence

of the D1 antagonist (P < 0.01, n = 5), when comparedwith their corresponding controls taken just prior to DAapplication (Supplemental Fig. S4B1). However, the D1antagonist SCH significantly reduced the effect of DA, as

Figure 3. Selective DA receptor stimulationA, pyramidal neuron filled with CG-1. The drawing marks position of a glass pipette filled with D1 receptor agonist,SKF38393 (500 μM). B, in the control sweep (Ctrl) an AP signal from the soma (ROI 0) is aligned with dendriticrecordings from 4 ROIs (1–4) shown in A. Black dots mark AP-mediated Ca2+ transients that were reduced inresponse to SKF. The effect of SKF was not detected 1 min or 3 min, after the cessation of the SKF pulse (washout).C, changes in Ca2+ signal amplitude in response to D1 receptor agonists SKF (50 μM), SKF (500 μM) or A-68930(A6) (500 μM). n indicates number of cells in data group. D, same as in C, except application pipette was filledwith D2 receptor agonist quinpirole (QP) at 500 μM and 5000 μM concentrations. E, the ability of DA to reduce thedendritic Ca2+ transient was established (3 open columns) before SCH was introduced in the bath and DA puffs(IPC, 10 mM) were repeated in the presence of SCH (3 grey columns). F, first, the DA effect on dendritic calciumwas established as explained in Fig. 1. Then, the patch pipette was removed and replaced with the one filled withinhibitors of protein kinases (Rp-cAMP, 500 μM; staurosporine, 0.4 μM). DA stimulus was repeated 30 min afterre-patching.



indicated by the comparison of the normalized data groups(Fig. 3E, compare the grey DA bar against the open DA bar,P < 0.01, n = 5). These results suggest the involvementof D1-like receptors in the rapid DAergic modulation ofdendritic Ca2+ (Figs 1 and 2).

Intracellular pathway – protein kinases

Dopamine is thought to exert its cellular effects via theactivation of protein kinases (Neve et al. 2004). In the nextseries of experiments we patched five neurons with normalintracellular solution (Regular IC) and established a potentDA effect on dendritic Ca2+ transients (Fig. 3F , opencolumn). Next, the original patch pipette was removed andreplaced by a second patch pipette, which was filled witha cocktail of protein kinase blockers (Rp-cAMP (500 μM)and staurosporine (0.4 μM)). After re-patching, the kinaseblockers were allowed to diffuse into the cytoplasm for atleast 30 min before repeating the DA stimulation on thesame dendrite. Re-patching of the neurons allowed us toperform a direct comparison of the DA effects on thesame cell, in the absence and in the presence of intra-cellularly applied drugs (Fig. 3F). A paired t test failedto detect statistical difference between the two conditions(P > 0.05, n = 5). These data suggest that the observedDA effects (Figs 1 and 2) were not mediated through theactivation or inactivation of intracellular protein kinases.

Dendritic AP waveforms

Two groups recently reported that DA suppresses dendriticCa2+, but they did not investigate if this result wasdue to a DA-induced perturbation of the AP amplitude(Day et al. 2008; Kisilevsky et al. 2008). In order toanalyse AP voltage waveforms in thin dendritic brancheswe employed a voltage-sensitive dye-imaging techniqueusing whole-field illumination (Acker & Antic, 2009)or laser-spot illumination (Zhou et al. 2007). DA waspressure applied from a glass micropipette positionednear the dendritic shaft (Fig. 4A). AP-associated dendriticvoltage signals were recorded before dopamine application(Fig. 4B, Ctrl), 2 s post-dopamine puff (DA), and againafter a 90 s washout period (Wash). In 7 out of 14neurons, local DA applications caused a small decreasein AP amplitude and a small increase in AP half-width(Fig. 4C). Group data showed that in apical dendrites(199 ± 67 μm from the soma) the average amplitude of thedendritic AP upon dopamine application compared withthe control values was 84.5 ± 3.7% (paired t test, P < 0.01,n = 14, N = 21; Fig. 4D). The AP half-width increased to106.1 ± 2.1% (P < 0.05; Fig. 4E), while the area under theAP did not change significantly (93.2 ± 4.7%, P > 0.05,Fig. 4F), compared with controls obtained before DAapplication.

We patched three apical dendrites (130–150 μm fromthe soma) and locally applied DA to the dendriticsegment where whole-cell patching was performed(Fig. 4G). Under infrared video-microscopy, we visuallymonitored the jet of ejected dopamine (IPC, 10 mM),while simultaneously recording dendritic AP voltage wave-forms (Fig. 4H). DA puffs (duration, 2 s) caused a smalldecrease in dendritic AP amplitude (98.0 ± 0.5%), anincrease in AP half-width (100.9 ± 1.1%), and a decreasein area under the AP (98.3 ± 1.1%) (13 DA applicationsin 3 neurons), compared with the control measurementsperformed just prior to DA application. A paired t testperformed on the data points before normalizationrevealed a statistically significant change in AP amplitude(P < 0.05), but not in AP half-width or area.

The next series of experiments addressed DAergicmodulation of AP waveform in the basal dendritesof PFC layer V pyramidal neurons (Fig. 5A) usingvoltage-sensitive dye imaging (Fig. 5B). DA applicationscaused small (∼5%), albeit statistically significant, changesin AP amplitude. The group data showed that in basaldendrites (165 ± 50.9 μm from the soma) the averageamplitude of dendritic APs upon dopamine application(IPC, 10 mM) was 94.8 ± 1.6% of the control values(paired t test, P < 0.01, n = 6; N = 19; Fig. 5C). TheAP half-width did not change (97.4 ± 4.2%, P > 0.05;Fig. 4D), while the area under AP decreased to 92.1 ± 2.1%(P < 0.01; Fig. 4E) compared with controls obtainedbefore DA application.

In summary, local DA stimuli (IPC, 10 mM) inducedsmall changes in the amplitudes of backpropagatingAPs in both apical and basal dendrites: ∼15% and∼5%, respectively. These data do not fit well withDA-induced changes in dendritic Ca2+ influx, whichwere approximately 40% (Fig. 1D, at IPC of 10 mM).Calcium channels are predominantly activated during therepolarization of the AP and therefore AP amplitude haslittle impact on Ca2+ influx. However, the shape of theAP in our experiments was minimally affected by the DApulse, according to the small changes in AP half-widthand AP area under the curve from both apical (Fig. 4Eand F) and basal dendrites (Fig. 5D and E). A substantialfraction of cortical pyramidal neurons do not respondto dopaminergic stimulation (Sesack & Bunney, 1989;Kisilevsky et al. 2008; Moore et al. 2011; present study).Even a small number of DA-insensitive neurons in thegroup data (Fig. 4E and F or Fig. 5D and E) would explainthe absence of a significant change in AP voltage wave-form. To improve the robustness and reliability of ourstudy we examined dendritic AP waveforms in cells thatwere positively responsive to DA stimulus. Towards thisgoal we employed dual-mode dendritic imaging (Fig. 6Aand B), where both Ca2+ and voltage transients wererecorded from the same region of interest (Canepariet al. 2007; Milojkovic et al. 2007). Calcium–voltage



imaging experiments performed in nine DA-responsivelayer V pyramidal cells showed a statistically significantreduction in AP-induced Ca2+ signal (55.9 ± 7.2% ofcontrol; P < 0.01, paired t test on raw data; Fig. 6E),which was not accompanied by a comparable changein AP amplitude (voltage-sensitive dye signal) in thesame dendritic segment (92.0 ± 4.1% of control; P < 0.05,paired t test on raw data; Fig. 6E). Next, we comparednormalized calcium signals against normalized voltagesignals obtained in the same ROI. Both calcium and voltagesignals were divided by their corresponding controlsobtained prior to DA application (Fig. 6C and D, Ctrl). ADA-induced decrease in the Ca2+ signal was significantlystronger than the DA-induced decrease in the voltage

signal in the same dendritic segment, as determinedby a paired t test on normalized data (P < 0.01, n = 9,Fig. 6F). Together, dendritic voltage imaging (Figs 4 and5), dendritic whole-cell recording (Fig. 4G and H) anddendritic voltage–calcium imaging (Fig. 6) suggest thatDA-induced suppression of AP-mediated dendritic Ca2+

influx is not due to reduction in dendritic AP amplitude.

Apical vs. basal dendrites

In four pyramidal neurons we were able to move theDA application pipette from a basal dendrite to theapical dendrite (Fig. 7A) and vice versa, without disrupting

Figure 4. Dendritic AP waveform upon DA stimulus – apical dendritesA, pyramidal neuron filled with voltage-sensitive dye JPW-3028. The drawing marks position of DA-filled pipette.B, dendritic AP waveforms before (Control), upon (DA) and 2 min after the DA application (Wash). Controlmeasurements were performed 3 times (Ctrl 1–3), while DA and Wash recordings were performed two timeseach. Each signal is temporal average of 3 sweeps. C, AP waveform Ctrl-3 is superimposed on AP waveformDA-1 to show DA-induced changes in AP amplitude and half-width. D, AP amplitudes during treatment withDA were normalized against the control measurements made before DA puff, and averaged across 14 apicaldendrites (N = 22 DA applications). E and F, same as D except AP half-width and AP area under the curve werequantified. Paired t tests were performed on raw data before normalization. G, IR DIC photograph of a patchedapical dendrite. H, dendritic action potentials before (Control) and during DA pressure ejections (DA-1 to DA-3).



the patch. An identical experimental protocol involvingdendritic Ca2+ imaging and local DA application wascarried out in both locations (Fig. 7B). On average, theDA-induced signal reduction (relative change in signalamplitude compared with control measurements) wasstronger in basal branches than that in the apical trunkof the same cell. A paired t test showed a statisticallysignificant difference in response between basal and apicaldendrites belonging to the same neuron (P < 0.05, n = 4,Fig. 7C, grey columns).

Next, we pooled the data from neurons treated withlocal application of DA (IPC, 10 mM) on either basal(N = 17) or apical (N = 8) dendrites. DA treatmentscaused 39.9 ± 2.1% reduction of Ca2+ signal in basaldendrites and 30.8 ± 3.9% reduction in apical dendrites(Fig. 7C, open columns). An unpaired t test indicatedthat in basal dendrites DA exerts slightly stronger, albeitstatistically significant (P < 0.05), inhibitory effects thanin the apical dendrites of the same neuron type.

Voltage-gated calcium channels

Availability and robustness of drugs that block L-,R-and T-type voltage-gated calcium channels (VGCC)

allowed us to ask if the DA effects on AP-inducedCa2+ transients were still present after the L-, R- andT-type channels were pharmacologically blocked. In thisexperimental design the inhibitory effect of DA puff wasfirst established in regular saline solution, by performingcontrol, DA1 and wash measurements (Fig. 8B, NormalACSF). Next, a cocktail of L-type calcium channelantagonists (verapamil (10 μM) and diltiazem (10 μM))were introduced in the bath. Addition of the cocktail(V+D) reduced the AP-associated dendritic Ca2+ trans-ients significantly (P < 0.01, n = 4, N = 7, Fig. 8C, left).Despite this drug-induced reduction in Ca2+ signal, atransient puff of DA was able to further reduce theamplitude of the dendritic signal (Fig. 8B, DA2). Therelative magnitude of the DA effect was still around 35%,regardless of blocking the L-type channel with V+D. TheDA2/Ctrl2 ratio obtained in drugs was not statisticallydifferent to the DA1/Ctrl1ratio obtained in normalACSF (P > 0.05, n = 4, N = 7, Fig. 8C, right). Nearlyidentical data were obtained when nickel (100–200 μM)was used to block R- and T-type VGCCs. Just as V+D,nickel also significantly reduced dendritic Ca2+ transients(P < 0.01, n = 5, N = 11, Fig. 8D, left). Despite blockingnickel-sensitive calcium conductances, a transient puff of

Figure 5. Dendritic AP waveform upon DA stimulus – basal dendritesA, pyramidal neuron filled with JPW-3028. Box marks the region targeted by a stationary laser beam. Inset, anactual image of this dendrite during ‘laser-spot’ illumination. B, voltage waveforms of dendritic action potentialsbefore (Control), upon DA pressure ejection (DA puff), and 90 s after DA application (Wash), recorded insidethe ROI. C–E, quantification of AP amplitude, AP half-width and AP area under the curve in 6 neurons (16 DAapplications on basal dendrites). Paired t tests were performed on raw data before normalization.



Figure 6. Calcium imaging and voltage imaging in the same dendriteOne pyramidal neuron filled with two fluorescent dyes. A, image captured using optical filters for bis-fura-2. B,image captured using filters for JPW-3028. Drawing marks position of the DA-filled pipette (IPC, 10 mM). C, somaticaction potential (whole-cell) is aligned with dendritic Ca2+ transients measured inside the box shown in A, before(Control), 4 s after DA puff and 2 min after the DA puff (Wash). D, dendritic action potential voltage waveformsmeasured inside the box shown in B, before (Control), upon DA puff, and 2 min after the DA puff (Wash). E, left,averaged Ca2+ signal amplitude (�F/F) in three conditions. Right, changes in voltage signal amplitude in the sameROI where Ca2+ signal was measured (n = 9). Paired t tests performed on raw data, before normalization. F, samedata as in E, except signal amplitudes were normalized in respect to corresponding control measurements. Pairedt test performed on the normalized data.

Figure 7. Apical vs. basal dendritesA, pyramidal neuron filled with OGB-1. Drawings mark two positions of the same DA-filled pipette. B, AP-associatedCa2+ transients were recorded in basal dendrite before (Control), upon DA pressure application (Dopamine) and1–2 min after DA treatment (Washout). The same DA pipette was then moved from basal dendrite onto the apicaltrunk and the same experimental protocol was repeated in the apical location (Apical row). C, relative reduction indendritic Ca2+ signal was averaged and plotted for two locations on the same neuron (n = 4 cells, grey columns,paired t test) and for a set of basal (N = 17) and set of apical dendrites (N = 8) belonging to different neurons(open columns, unpaired t test). All dendrites in this figure were treated with the same IPC of DA (10 mM).



DA was still able to further reduce the amplitude of thedendritic calcium signal. The DA2/Ctrl2 ratio obtained innickel was not statistically different than the DA1/Ctrl1ratio obtained in regular ACSF (P > 0.05, n = 5, N = 11,Fig. 8D, right). These data suggest that the observedDA effect on AP-associated dendritic Ca2+ transients(Figs 1–7) was not primarily based on inhibition of L-,R-or T-type VGCCs.

GABA vs. dopamine

So far, our experimental data suggest that DAergicmodulation interferes with some, but not all, aspects ofdendritic AP backpropagation. For example, dendriticCa2+ influx was markedly reduced (Fig. 1) but thedendritic AP voltage waveform was not strongly alteredby DA (Fig. 6). GABA has been shown to inhibit dendriticelectrical and calcium transients (Tsubokawa & Ross,1996; Larkum et al. 1999; Couey et al. 2007; Vogt et al.2011); thus, we wished to compare GABAergic and

DAergic modulation of AP backpropagation in the PFCcortical pyramidal neurons. To provide the most directcomparison between these two neurotransmitters, in thenext series of experiments we applied sequential puffsof GABA and DA onto the same dendritic segment,while measuring dendritic AP-associated Ca2+ transientsalong the entire branch. In this experimental protocoltwo glass pipettes (one filled with GABA (1 mM) and theother filled with DA (10 mM)) were both positioned inthe mid-dendritic segment, as shown in the schematicdrawing in Fig. 9A. Local puffs of GABA (duration,20–200 ms) caused elimination of AP-mediated Ca2+

influx distally from the drug application site (Fig. 9B,arrow). Local puffs of DA (duration, 500 ms), on theother hand, caused only local reductions in Ca2+ signalat the sites of DA application (Fig. 9C ROI 2), whileCa2+ transients in dendritic segments distal to the DAapplication site were unaffected (Fig. 9C, arrow). Thissuggests a successful AP backpropagation through thesite of DA application. The actual image of a neuronand additional recording sites during sequential GABA

Figure 8. DA effect in the absence of L-type and R/T-type calcium channelsA, pyramidal neuron filled with OGB-1. Drawing marks the position of the DA-filled pipette (IPC, 10 mM). B,AP-associated Ca2+ transients were recorded in basal dendrite before (Ctrl1), upon DA pressure application (DA1),and 1–2 min after DA treatment (Wash1). Next, a cocktail of L-type calcium channel antagonists (verapamil, 10 μM)and diltiazem (10 μM), ‘V+D’) were added to the perfusate, and the same experimental protocol was repeated. C,left, average amplitudes of control Ca2+ transients before (ACSF) and after addition of V+D, expressed as �F/F.The cocktail ‘V+D’ potently reduced dendritic AP-Ca2+ transients. Paired t test on raw data, ∗∗P < 0.01, n = 4,N = 7. Right, the DA effect, expressed as a relative reduction in dendritic Ca2+ signal, is shown before (ACSF)and after introduction of V+D in the same cell. Paired t test, †P > 0.05, n = 4, N = 7. D, same as in C exceptbrain slices were perfused with R/T-type channel antagonist nickel (100–200 μM). ∗∗P < 0.01; †P > 0.05, n = 5,N = 11.



and DA applications are displayed in SupplementalFig. S5. The group data obtained in five neurons showedthat application of GABA caused the elimination ofAP-associated Ca2+ transient distal to the application site(Fig. 9D, +80 μm), while DA-induced changes (n = 7)were restricted to the DA application site (Fig. 9E, DA,0 μm).

Dendritic voltage waveforms upon GABA stimulation

Elimination of AP-mediated Ca2+ influx distal to theGABA application site (Fig. 9B, ROI 3) could have beendue to a complete block of AP backpropagation, or due toGABA modulation of Ca2+ flux. To resolve this questionwe performed multisite voltage-sensitive dye recordingsof backpropagating APs using whole-field illumination(Acker & Antic, 2009). Amplitudes of dendritic APswere severely attenuated at the GABA application sites(Fig. 10Ab, Sweep 2, ROI 2a), but also, and to an evengreater extent, in the dendritic segments distal to the

GABA application site (ROIs 2b and 2c). Identical datawere obtained in three layer V pyramidal neurons.

It is currently thought that GABA depolarizes dendriticbranches of cortical pyramidal neurons (Gulledge &Stuart, 2003). The depolarizing action of GABA mayamplify AP-associated voltage transients in distal dendriticbranches, rather than attenuate them as observed inour experiments (Fig. 10A). Experiments described inFig. 10A were performed with patch pipettes in whole-cellconfiguration. The ensuing dialysis of the neuronalintracellular space disrupts the natural concentration ofchloride ions (Cl−) and the GABAA receptor reversalpotential (Kaila, 1994). To address this caveat, in thenext series of experiments we analysed the GABAergiceffect on dendritic voltage transients without the use ofpatch electrodes (Canepari et al. 2010). Neurons werefilled with the voltage-sensitive dye and then dye-loadingpipettes were removed. Canepari et al. (2010) have shownthat physiological concentrations of Cl− are establishedwithin 10–15 min of removal of the patch pipette. Weallowed more than 60 min between pipette removal and

Figure 9. DA vs. GABA on the same dendriteA, experimental outline showing DA-filled and GABA-filled pipette converging on a middle segment of one basaldendrite. B, in the first sweep (Control) the cell body was injected with short current pulses to produce a train ofsomatic APs (ROI 0). The corresponding dendritic Ca2+ transients were recorded simultaneously from 3 ROIs (1–3)and arbitrarily scaled for display. In the next sweep (GABA) APs were paired with GABA puff. Control and GABAsweeps are superimposed for comparison. C, same cell, same detectors, same scaling as in B except recordingswere made before (Control) and after local application of dopamine (DA). D and E, drug-induced changes in signalamplitude along a dendritic branch normalized against control measurements in neurons treated either with GABA(n = 5) or DA (n = 7) on the mid-segment of a basal branch, as depicted schematically below each graph. FifthAP in the train was used for quantification of these data. Control measurements were obtained just prior to eachdrug application. GABA: IPC, 1 mM; duration, 20–200 ms. DA: IPC, 10 mM; duration, 500 ms. Local application ofDA causes a spatially restricted reduction in dendritic AP-mediated Ca2+ influx. DA application at mid-dendriticsegment did not affect AP signals in distal dendritic tip (arrow in C). GABA puff eliminated AP-Ca2+ signals fromdendritic segments distally from the GABA application site (arrow in B).



the beginning of the optical recording session. Withintact ionic milieu (without patch electrode, n = 3), thesuppressing effect of GABA on the AP amplitude wastransient, localized to the target branch (Fig. 10Ba–bdendrite 2), and removed by washing (Fig. 10Bb, sweeps 3and 5). Quantification of group data (Fig. 10Bc) showedthat local GABA puffs caused a clear block in AP back-

propagation in both groups of neurons: Group 1 cellswith the patch pipette attached in whole-cell configuration(open) and Group 2 cells free from the patch pipette (grey).Unlike dopamine (Fig. 9C, arrow), the local GABAergicapplications obstructed propagation of electrical trans-ients along the GABA-receiving dendritic branch (Figs 9Dand 10).

Figure 10. Effect of GABA on dendritic AP waveforms – neurons with intact intracellular [Cl−]Aa, pyramidal neuron filled with JPW-3028. Fluorescent image inverted. Ab, in the control sweep (Control) acurrent-evoked AP is recorded simultaneously via the patch pipette (ROI 0) and optically at 6 ROIs shown in Aa.In the next sweep (GABA) a puff of GABA is delivered at the location marked by drawing in Aa. GABAergicstimulus suppresses dendritic voltage transients locally (ROI 2a), but also in distal dendritic segments (ROIs 2b and2c). The effect of GABA was reversed 90 s later (Wash). Ba, pyramidal neuron was filled with JPW-3028 and thedye-loading patch pipette was removed. Bb, in the control sweep (Control) an AP was evoked by extracellularelectric shock and recorded simultaneously at 5 ROIs shown in Ba. In the next sweep (sweep 2 GABA) a puff ofGABA is delivered at the location marked by drawing in Ba. The effect of GABA is reversed 90 s later (Sweep 3Wash). The second GABA application and its wash (sweeps 4 and 5) were similar to the first application and itswash (sweeps 2 and 3). Bc, peak amplitudes of dendritic AP voltage waveforms obtained during GABA application(sweep 2) were normalized to control measurements (sweep 1), binned across 50 μm-long dendritic segments,and plotted against the relative distance from the GABA application site. Regardless of the presence (open) orabsence (grey) of the whole-cell pipette, GABA causes a complete block of AP backpropagation distally from theapplication site (positive micrometer values).



Physiological changes during dopamine modulationof dendritic calcium influx

The physiological significance of dopaminergicsuppression of dendritic Ca2+ (Figs 1–7) was studied nextby monitoring two parameters: (a) the excitability of theneuronal cell body, and (b) the frequency-dependentAP backpropagation in basal dendrites; during the timeperiod when DA stimulus was presented to the basaldendrite.

(a) Neuronal excitability – somatic action potential firing.Generation of a sodium AP in the axon initial segment(Palmer & Stuart, 2006; Popovic et al. 2011) is the primarygoal of synaptic integration. We asked if the ability ofa pyramidal cell to generate APs was changed duringDAergic modulation of dendritic function (Figs 1 and 2).In this series of experiments we first used calcium imaging(Fig. 11A) to determine whether dendritic calcium trans-ients were sensitive to DAergic modulation (Fig. 11B), andonly in this case was the experiment continued. Next,we injected a depolarizing current pulse (C.I., duration,800 ms) into the cell body to generate a train of APs(Fig. 11C, Ctrl). The same DAergic stimulus, previouslyshown to suppress dendritic Ca2+ influx (Fig. 11B, DA),was then re-applied in the middle of the AP train in anattempt to alter the rate of neuronal AP firing (Fig. 11C,DA). APs were counted in the time window marked bythe rectangle (Fig. 11C). A paired t test failed to detectany significant differences between sweeps acquired beforeDA application (Fig. 11D, Ctrl) and those acquired duringDA application (DA, paired t test, P > 0.05, n = 7). Thesedata indicate that dopaminergic stimuli received in distal

dendritic segments do not affect current-evoked APs inthe soma.

(b) Frequency-dependent AP backpropagation. It waspreviously shown that Ca2+ influx in basal dendrites ofneocortical pyramidal neurons is frequency dependent.Robust Ca2+ signals were detected only at high firingrates (Kampa et al. 2006). In light of the importance ofhigh-frequency AP bursts for synaptic plasticity in basaldendrites (Kampa et al. 2006) we designed experimentsto test if DAergic regulation of AP-associated Ca2+ signalsin basal dendrites was frequency dependent (Fig. 12A).In control measurements, pyramidal neurons were drivenby brief current pulses to fire three APs, while dendriticCa2+ signals were measured along the basal dendrite(Fig. 12B, Control). In the following trials, iontophoreticDA pulses (duration, 300 ms; intensity, 1.2 μA) wereapplied 400 ms before AP triplets (command pulses shownin Fig. 12A, top). Dopaminergic stimulation of the basaldendrite (160.4 ± 12.2 μM to the soma, n = 7) did notcause any obvious changes in AP backpropagation atvarious frequencies, apart from a simple scaling-down ofthe calcium signal amplitude (Fig. 12C). This could bedue to one of two reasons. First, non-linear phenomenasuch as AP-induced dendritic Ca2+ electrogenesis areusually found in dendrites at large distances from thecell body (Larkum et al. 1999; Kampa & Stuart, 2006).Second, the intensity of the DA stimulus was perhapstoo weak to reach a critical DA concentration requiredto alter frequency-dependent AP backpropagation. Toaddress these two issues we tested more distal segmentsof basal dendrites (179.8 ± 13.6 μm, n = 4) using a 50%stronger DA stimulus (1.8 μA). The stronger DA stimuluscaused stronger reductions in the Ca2+ signal amplitude

Figure 11. Dendritic stimulation with DA does not affect somatic excitabilityA, pyramidal neuron filled with OGB-1. The drawings mark the positions of sharp (40 M�) DA-filled pipette (IPC,100 mM). B, AP-associated dendritic Ca2+ transients (dend.) were recorded before (dashed grey line) and afterDA treatment (black line) and superimposed for comparison. Exact timing of DA iontophoretic pulse (duration,300 ms) is marked by the bottom trace. C, same DA puff was applied while the neuron was driven by currentinjection (C.I.; duration, 800 ms) to fire APs. Rectangle indicates time window used to quantify AP frequency. D,average AP firing frequency before (Ctrl) and during DA application on basal dendrite (DA).



(Fig. 12D). The DA effect again was uniform acrossall frequencies within the tested range, as normalizedresponses revealed identical curves in Control and DAconditions (Fig. 12C and D, ratio 3 AP/1 AP). In summary,although the amplitude of the Ca2+ signal was strongly

reduced by DA (Fig. 12CD, upper graphs), the relativeincrease in dendritic Ca2+ signal in response to growing APfrequency was not affected by DA (Fig. 12C and D, lowergraphs). This shows that the effect of DA was uniformacross different AP firing frequencies.

Figure 12. DA effect on dendritic calcium during high-frequency AP firingA, pyramidal neuron filled with OGB-1. Drawings marks the position of DA-filled pipette (IPC, 10 mM) at 145 μM

path distance from the centre of the soma. B, current-injection evoked AP triplets were recorded electrically fromthe cell body (soma) and optically from ROI shown in A (dend). Current injection and DA iontophoresis commandsare also shown in A. Recordings made before DA application (Control, dashed grey line) and upon DA iontophoresis(DA, black line) are superimposed for comparison. In the same neuron, the Control and DA recordings were madeacross a range of AP firing rates (8.3–125 Hz), as indicated. C, upper graph, for a given AP firing rate the peakamplitude of dendritic Ca2+ signals (expressed as �F/F) was averaged across 7 neurons before (Control) and aftertreatment with iontophoretic current intensity of 1.2 μA (Dopamine). An average distance from the soma forROIs was 160.4 ± 12.2 μM (n = 7). Lower graph, relative increase in Ca2+ signal amplitude with increasing APfiring frequency. Inset, first AP at 8.3 Hz was used for calculating ratio 3AP /1AP. An example of 3 AP (50 Hz) isshown on the right. D, same as in C, except DA iontophoretic current was set to 1.8 μA and ROI was more distal(179.8 ± 13.6 μm, n = 4).



Discussion

Our current experiments attempted to mimic the spatialand temporal attributes of synaptic DAergic stimuli thatimpinge on dendrites of cortical pyramidal neurons(Goldman-Rakic et al. 1989) during phasic DA signalling(Ljungberg et al. 1992) by application of exogenous DAdirectly onto a dendritic branch (Day et al. 2008). Wefound that brief puffs of DA, delivered either by pressure(Figs 1, 6 and 7), or iontophoretic current (Figs 2, 11and 12), transiently suppressed AP-mediated dendriticCa2+ influx. The DA effect was rapid, as changes in theamplitude of Ca2+ influx were regularly detected less than500 ms from the onset of DA pulse (Fig. 2B), consistentwith the time scale of a 500 ms (phasic) burst of midbraindopaminergic neurons that project into the prefrontalcortex (Ljungberg et al. 1992; Schultz, 2002). The DA effectwas mimicked by D1-like receptor agonists, independentfrom the activation of intracellular protein kinases, andwas exclusively restricted to the dendritic membrane indirect contact with a high concentration of DA molecules.These results suggest that DA molecules interfere withcalcium influx via a direct protein/G-protein link betweenthe D1 receptor and voltage-gated Ca2+ channel (Dolphin,2003; Kisilevsky et al. 2008), or by clogging the pore ofthe calcium channel, or both. AP voltage waveforms inthe basal dendrites of PFC neurons are very sensitiveto perturbations in voltage-gated Na+ and A-type K+

current (Acker & Antic, 2009). DA molecules specificallyaffected VGCCs, but to a far lesser degree the dendriticNa+ channels or K+ channels, as relatively strong DAergicstimuli had a small effect on the dendritic AP wave-form (Figs 4 and 5). Dual voltage–calcium imaging data(Fig. 6) eliminated a possibility that DA-resistant neurons,‘non-responders’ (Sesack & Bunney, 1989; Surmeier et al.1995; Day et al. 2008; Kisilevsky et al. 2008; Mooreet al. 2011), contaminated our experimental groups andbiased our data interpretation toward the conclusion thatDA stimulation minimally affects dendritic AP wave-form. That the DA effect on the AP voltage waveformwas negligible was also supported by a large number ofmeasurements that documented successful propagation ofAP through the DA application site. That is, AP-associatedCa2+ transients were regularly found distally from the DAapplication site (Figs 2A and 9E).

Voltage-gated calcium channels

Taken together, our data indicate that the primary effectof a brief dopaminergic stimulus is a temporary inhibitionof dendritic VGCCs. The precise identity (subtype) ofthe dendritic Ca2+ channel involved is unclear. In corticallayer V pyramidal neurons, an AP propagating from thesoma into the apical dendrite evokes a rapid Ca2+ trans-ient via activation of four VGCC subtypes: L-, N-, P-

and R-type, each contributing to the calcium influx 25%,28%, 10% and 37%, respectively (Markram et al. 1995).The detection of these four VGCC subtypes is consistentwith the findings from other groups (Westenbroek et al.1992; Lorenzon & Foehring, 1995; Castelli & Magistretti,2006). T-type VGCCs were detected in dendrites byhistochemistry (Craig et al. 1999; McKay et al. 2006) butnot confirmed by electrophysiology (Almog & Korngreen,2009). Therefore, there are at least six potential VGCCsubtypes populating cortical pyramidal neurons: L-, N-,P-, Q-, R- and T-type. It is experimentally challengingto determine if only one, or two or more VGCC sub-types are suppressed by DA. Some experimental dataare already available to help resolve this issue. First, ourexperiments with VGCC blockers (Fig. 8) rejected thepossibility that L-, R- and T-type channels are primarysubstrates of the observed DA effect. Second, N-typechannels in dendrites of rat PFC neurons are stronglyinhibited by DA through a D1 receptor–N-type channelsignalling complex (Kisilevsky et al. 2008). In mediumspiny neurons, the presence of the D2 receptor agonist,quinpirole, reduces AP-evoked Ca2+ influx through R-typevoltage-gated Ca2+ channels (Higley & Sabatini, 2010).However, another group have shown that applicationsof D1 agonists transiently reduced N- and P-type Ca2+

currents in medium spiny neurons (Surmeier et al. 1995).Therefore, several lines of arguments support the idea thatN-, P- and Q-type channels are likely to be the targetsof DAergic synaptic inputs on PFC pyramidal neurons(Goldman-Rakic et al. 1989).

Dopamine concentration

Ejection of drugs from micropipettes (∼7 M�) typicallyresults in extracellular drug concentrations that are oneor two orders of magnitude lower than that insidethe pipette. Still, one may argue that the extracellularDA concentrations achieved in the present study wereunphysiological. The exact physiological range of DAconcentrations generated inside the synaptic cleft upona phasic discharge of an assembly of midbrain neurons(Grace & Bunney, 1984; Ljungberg et al. 1992) iscurrently unknown. The concentration of extrasynapticDA, measured by a large carbon fibre (or microdialysis), isa severe underestimate of actual DA concentration peaksin the extracellular space due to a large surface area of theactive electrode and the scarcity of DA-releasing boutons.Furthermore, the reported micromolar concentrations ofDA in the PFC extracellular space (Garris & Wightman,1994) are very different from the actual peak DAconcentration inside the DA synaptic cleft sealed by glialprocesses. Some researchers postulate that peaks of DAconcentration inside synaptic clefts of DA synapses mayreach millimolar values during an outburst of midbrain



DA neuron firing (Garris et al. 1994; Grace et al. 2007).The present results and conclusions (Figs 1–8) pertain todendritic plasmalemma directly opposed to DA synapses.

GABA vs. dopamine

GABAergic and DAergic axons make direct synapticcontacts on dendritic branches of cortical pyramidalneurons (Goldman-Rakic et al. 1989; Somogyi et al. 1998).In our experimental protocol (local puff in vitro) bothneurotransmitters suppress AP-associated dendritic Ca2+

influx. Simultaneous multisite measurements revealedan interesting spatio-temporal distinction between thesetwo forms of inhibition. GABAergic stimuli unselectivelyblocked both dendritic Ca2+ and dendritic electricalsignals distally from the GABA application sites (Figs 9Dand Fig. 10). Local applications of DA, on the otherhand, suppressed dendritic Ca2+ transients only locally,but they did not block sodium APs that pass throughthe site of DA input on the way to distal dendriticsegments (Fig. 9E). Our summary figure (SupplementalFig. S6) is consistent with anatomical organizations of thecortical GABAergic and DAergic systems. GABA-releasingfibres impinge on dendritic shafts (Beaulieu & Colonnier,1985), while DA fibres mostly target individual dendriticspines (Goldman-Rakic et al. 1989). A transient releaseof GABA provides robust and reliable inhibition of anentire segment of the dendritic tree distally from the GABAinput (Supplemental Fig. S6A), while DAergic inhibitionis more selective, probably confined to a few dendriticspines (Hoogland & Saggau, 2004), less potent than theGABAergic inhibition (Gulledge & Stuart, 2003; Kisilevskyet al. 2008; Higley & Sabatini, 2010), and does notrepresent an obstacle for propagation of informationencoded by electrical transients – APs (SupplementalFig. S6B).

Our data suggest that the role of the synapticdopamine release is to reduce calcium influx in thepostsynaptic compartment, without interfering with thepropagation of electrical signals. Two major corticalneurotransmitters (glutamate and GABA) concurrentlychange both electrical and calcium signals. Dopamine isunique because it can modulate the calcium flux withoutaffecting electrical coupling between two compartments.An electrical signal of sufficient amplitude (e.g. AP) canpass the dopamine input site and continue its propagationalong the dendrite, with no apparent loss of information.

Other functional implications

Our data (Fig. 11) indicate that local dopaminergicmodulation of Ca2+ influx in distal dendrites does notinterfere with the AP initiation, which takes place inthe axon initial segment (Kole et al. 2008; Popovic

et al. 2011). Rather, the effect of a brief and local(synaptic-like) DA input is exclusively restricted to onedendritic segment engulfed in DA. This is consistentwith the anatomical distribution of DA synapses in distaldendrites (Goldman-Rakic et al. 1989). It is also consistentwith the spatially restricted effect on AP-associated Ca2+

influx (Fig. 2A), and with the finding that cytosolic proteinkinases are not involved (Fig. 3F). In other words, brief DAstimuli received on one dendritic segment do not producediffusible intracellular signals capable of changing thephysiology of adjacent neuronal compartments (Figs 2Aand 11). However, this is not to say that dopaminergicafferents in the PFC are completely uncoupled from theAP initiation process. Given the recent evidence of theselective inhibition of T-type calcium channels in theaxon initial segment by dopamine, leading to markedAP firing pattern changes (Bender et al. 2012), as wellas the documented high density of dopaminergic fibresin deep cortical lamina, where axon initial segments oflayer V pyramidal neurons reside (Sesack et al. 1989), wecannot entirely exclude that DA release from midbrainfibres in the deep lamina of the PFC will also impact onthe AP initiation process. In fact, in PFC layer V pyramidalneurons a brief DA application delivered approximately30 μm from the axon initial segment (at the neuronal cellbody), causes a severe interruption in AP firing (Mooreet al. 2011, their Fig. 4).

Suppression of synaptic plasticity may be one functionalrole of synaptic dopamine. Based on the importance ofintracellular Ca2+ (Sabatini et al. 2001) and a robustDA-mediated suppression of dendritic Ca2+ (Figs 1–8),it is possible that phasic dopamine activity affectsthe plasticity of pyramidal neurons in the prefrontalcortex. For example, during phasic AP firing in mid-brain dopaminergic neurons, the dendritic spines andshort segments of dendritic shafts near DA contacts(Supplemental Fig. S6B, red sphere) receive less Ca2+

and therefore would be excluded from Ca2+-mediatedprocesses, such as synaptic plasticity. In this scenario, aphasic DAergic signal (Grace & Bunney, 1984; Ljungberget al. 1992) would serve to protect a subset of synapticinputs from undergoing modulations in efficacy. Onany given pyramidal cell, only a fraction of dendriticspines receives direct synaptic inputs from DAergic fibres(Bergson et al. 1995). One is tempted to postulatethat dendritic spines that do receive direct synaptic DAcontacts are ‘strategically important’. By being essentialfor the nature of the neuron’s computational task,these strategically important synapses are targeted bydopaminergic axon terminals. The synaptic efficacy ofsuch synapses is preserved in the long run, as they resistCa2+-mediated functional changes by receiving DA duringsalient events, such as pleasure, stress, punishment andsexual arousal (Abercrombie et al. 1989; Schultz, 2002).The proposed scenario does not exclude the role of DA



in promoting synaptic plasticity. At the sites of directDAergic contact the synaptic plasticity is opposed by highconcentration of DA, but DA spill-over from DAergicsynapses, as well as dopamine released from non-synapticDAergic boutons, operate in low concentration regimenand as such may promote synaptic plasticity (Jay, 2003;Seamans & Yang, 2004; Matsuda et al. 2006).

Protection from excitotoxicity may be one of thefunctional roles attributable to synaptic dopamine. Thishypothesis is derived from the harmful effect of Ca2+

accumulation in the intracellular cytosol (Dirnagl et al.1999). Each layer V pyramidal cell in the mammalianPFC receives ∼15,000 anatomically distinct glutamatergiccontacts (Elston, 2003), whose activation brings massiveloads of Ca2+ into the cell (Milojkovic et al. 2007).Although the exact number of simultaneously activerelease sites is difficult to determine, a large number ofexcitatory glutamatergic vesicles impinge on the dendriticmembrane every second of an intense cortical process.Given the hourly duration of some cortical tasks, one mustrespect the normal physiological mechanisms employedto protect dendrites from Ca2+ overload (Sapolsky, 2001).We hypothesize that synaptic dopamine release on distaldendritic segments may play a role in this process. Duringintense cortical activity dopaminergic inputs to PFC mayserve to prevent local Ca2+ overload and avert injury(Arundine & Tymianski, 2003). Interestingly, in peoplesuffering from schizophrenia there is reduced amount ofdopamine (Abi-Dargham & Moore, 2003) and less brainactivity in the PFC (Meyer-Lindenberg et al. 2002) –hypodopaminergia and hypofrontality. We speculate thatglutamate excitotoxicity, which has been implicated inthe schizophrenic hypofrontality (Stone et al. 2007), mayhave been the consequence of chronic prefrontal hypo-dopaminergia.

References

Abercrombie ED, Keefe KA, DiFrischia DS & Zigmond MJ(1989). Differential effect of stress on in vivo dopaminerelease in striatum, nucleus accumbens, and medial frontalcortex. J Neurochem 52, 1655–1658.

Abi-Dargham A & Moore H (2003). Prefrontal DAtransmission at D1 receptors and the pathology ofschizophrenia. Neuroscientist 9, 404–416.

Acker CD & Antic SD (2009). Quantitative assessment of thedistributions of membrane conductances involved in actionpotential backpropagation along basal dendrites. JNeurophysiol 101, 1524–1541.

Akerman SN, Zorec R, Cheek TR, Moreton RB, Berridge MJ &Mason WT (1991). Fura-2 imaging of thyrotropin-releasinghormone and dopamine effects on calcium homeostasis ofbovine lactotrophs. Endocrinology 129, 475–488.

Almog M & Korngreen A (2009). Characterization ofvoltage-gated Ca2+ conductances in layer 5 neocorticalpyramidal neurons from rats. PLoS One 4, e4841.

Antic SD (2003). Action potentials in basal and obliquedendrites of rat neocortical pyramidal neurons. J Physiol550, 35–50.

Arundine M & Tymianski M (2003). Molecular mechanisms ofcalcium-dependent neurodegeneration in excitotoxicity. CellCalcium 34, 325–337.

Beaulieu C & Colonnier M (1985). A laminar analysis of thenumber of round-asymmetrical and flat-symmetricalsynapses on spines, dendritic trunks, and cell bodies in area17 of the cat. J Comp Neurol 231, 180–189.

Bender KJ, Ford CP & Trussell LO (2010). Dopaminergicmodulation of axon initial segment calcium channelsregulates action potential initiation. Neuron 68, 500–511.

Bender KJ, Uebele VN, Renger JJ & Trussell LO (2012). Controlof firing patterns through modulation of axon initialsegment T-type calcium channels. J Physiol 590, 109–118.

Bergson C, Mrzljak L, Smiley JF, Pappy M, Levenson R &Goldman-Rakic PS (1995). Regional, cellular, and subcellularvariations in the distribution of D1 and D5 dopaminereceptors in primate brain. J Neurosci 15, 7821–7836.

Canepari M, Djurisic M & Zecevic D (2007). Dendritic signalsfrom the rat hippocampal CA1 pyramidal neurons duringcoincident pre- and post-synaptic activity: a combinedvoltage- and calcium-imaging study. J Physiol 580, 463–484.

Canepari M, Willadt S, Zecevic D & Vogt KE (2010). Imaginginhibitory synaptic potentials using voltage sensitive dyes.Biophys J 98, 2032–2040.

Castelli L & Magistretti J (2006). High-voltage-activated Ca2+currents show similar patterns of expression in stellate andpyramidal cells from rat entorhinal cortex layer II. Brain Res1090, 76–88.

Couey JJ, Meredith RM, Spijker S, Poorthuis RB, Smit AB,Brussaard AB & Mansvelder HD (2007). Distributednetwork actions by nicotine increase the threshold forspike-timing-dependent plasticity in prefrontal cortex.Neuron 54, 73–87.

Craig PJ, Beattie RE, Folly EA, Banerjee MD, Reeves MB,Priestley JV, Carney SL, Sher E, Perez-Reyes E & Volsen SG(1999). Distribution of the voltage-dependent calciumchannel α1G subunit mRNA and protein throughout themature rat brain. Eur J Neurosci 11, 2949–2964.

Day M, Wokosin D, Plotkin JL, Tian X & Surmeier DJ (2008).Differential excitability and modulation of striatal mediumspiny neuron dendrites. J Neurosci 28, 11603–11614.

Dirnagl U, Iadecola C & Moskowitz MA (1999). Pathobiologyof ischaemic stroke: an integrated view. Trends Neurosci 22,391–397.

Dolphin AC (2003). G protein modulation of voltage-gatedcalcium channels. Pharmacol Rev 55, 607–627.

Elston GN (2003). Cortex, cognition and the cell: New insightsinto the pyramidal neuron and prefrontal function. CerebCortex 13, 1124–1138.

Few WP, Scheuer T & Catterall WA (2007). Dopaminemodulation of neuronal Na+ channels requires binding of Akinase-anchoring protein 15 and PKA by a modified leucinezipper motif. Proc Natl Acad Sci U S A 104, 5187–5192.

Garris PA, Ciolkowski EL, Pastore P & Wightman RM (1994).Efflux of dopamine from the synaptic cleft in the nucleusaccumbens of the rat brain. J Neurosci 14, 6084–6093.



Garris PA & Wightman RM (1994). Different kinetics governdopaminergic transmission in the amygdala, prefrontalcortex, and striatum: an in vivo voltammetric study. JNeurosci 14, 442–450.

Goldman-Rakic PS, Leranth C, Williams SM, Mons N &Geffard M (1989). Dopamine synaptic complex withpyramidal neurons in primate cerebral cortex. Proc NatlAcad Sci U S A 86, 9015–9019.

Grace AA & Bunney BS (1984). The control of firing pattern innigral dopamine neurons: burst firing. J Neurosci 4,2877–2890.

Grace AA, Floresco SB, Goto Y & Lodge DJ (2007). Regulationof firing of dopaminergic neurons and control ofgoal-directed behaviors. Trends Neurosci 30, 220–227.

Gulledge AT & Stuart GJ (2003). Action potential initiation andpropagation in layer 5 pyramidal neurons of the ratprefrontal cortex: absence of dopamine modulation. JNeurosci 23, 11363–11372.

Higley MJ & Sabatini BL (2010). Competitive regulation ofsynaptic Ca2+ influx by D2 dopamine and A2A adenosinereceptors. Nat Neurosci 13, 958–966.

Hoffman DA & Johnston D (1999). Neuromodulation ofdendritic action potentials. J Neurophysiol 81, 408–411.

Hoogland TM & Saggau P (2004). Facilitation of L-type Ca2+channels in dendritic spines by activation of β2 adrenergicreceptors. J Neurosci 24, 8416–8427.

Jay TM (2003). Dopamine: a potential substrate for synapticplasticity and memory mechanisms. Prog Neurobiol 69,375–390.

Kaila K (1994). Ionic basis of GABAA receptor channelfunction in the nervous system. Prog Neurobiol 42, 489–537.

Kampa BM, Letzkus JJ & Stuart GJ (2006). Requirement ofdendritic calcium spikes for induction ofspike-timing-dependent synaptic plasticity. J Physiol 574,283–290.

Kampa BM & Stuart GJ (2006). Calcium spikes in basaldendrites of layer 5 pyramidal neurons during actionpotential bursts. J Neurosci 26, 7424–7432.

Kisilevsky AE, Mulligan SJ, Altier C, Iftinca MC, Varela D, TaiC, Chen L, Hameed S, Hamid J, Macvicar BA & ZamponiGW (2008). D1 receptors physically interact with N-typecalcium channels to regulate channel distribution anddendritic calcium entry. Neuron 58, 557–570.

Kole MH, Ilschner SU, Kampa BM, Williams SR, Ruben PC &Stuart GJ (2008). Action potential generation requires a highsodium channel density in the axon initial segment. NatNeurosci 11, 178–186.

Larkum ME, Kaiser KM & Sakmann B (1999). Calciumelectrogenesis in distal apical dendrites of layer 5 pyramidalcells at a critical frequency of back-propagating actionpotentials. Proc Natl Acad Sci U S A 96, 14600–14604.

Ljungberg T, Apicella P & Schultz W (1992). Responses ofmonkey dopamine neurons during learning of behavioralreactions. J Neurophysiol 67, 145–163.

Lledo PM, Legendre P, Israel JM & Vincent JD (1990).Dopamine inhibits two characterized voltage-dependentcalcium currents in identified rat lactotroph cells.Endocrinology 127, 990–1001.

Lorenzon NM & Foehring RC (1995). Characterization ofpharmacologically identified voltage-gated calcium channelcurrents in acutely isolated rat neocortical neurons. I. Adultneurons. J Neurophysiol 73, 1430–1442.

McKay BE, McRory JE, Molineux ML, Hamid J, Snutch TP,Zamponi GW & Turner RW (2006). CaV3 T-type calciumchannel isoforms differentially distribute to somatic anddendritic compartments in rat central neurons. Eur JNeurosci 24, 2581–2594.

Malgaroli A, Vallar L, Elahi FR, Pozzan T, Spada A & MeldolesiJ (1987). Dopamine inhibits cytosolic Ca2+ increases in ratlactotroph cells. Evidence of a dual mechanism of action. JBiol Chem 262, 13920–13927.

Markram H, Helm PJ & Sakmann B (1995). Dendritic calciumtransients evoked by single back-propagating actionpotentials in rat neocortical pyramidal neurons. J Physiol485, 1–20.

Matsuda Y, Marzo A & Otani S (2006). The presence ofbackground dopamine signal converts long-term synapticdepression to potentiation in rat prefrontal cortex. J Neurosci26, 4803–4810.

Maurice N, Tkatch T, Meisler M, Sprunger LK & Surmeier DJ(2001). D1/D5 dopamine receptor activation differentiallymodulates rapidly inactivating and persistent sodiumcurrents in prefrontal cortex pyramidal neurons. J Neurosci21, 2268–2277.

Meyer-Lindenberg A, Miletich RS, Kohn PD, Esposito G,Carson RE, Quarantelli M, Weinberger DR & Berman KF(2002). Reduced prefrontal activity predicts exaggeratedstriatal dopaminergic function in schizophrenia. NatNeurosci 5, 267–271.

Milojkovic BA, Zhou WL & Antic SD (2007). Voltage andcalcium transients in basal dendrites of the rat prefrontalcortex. J Physiol 585, 447–468.

Moore AR, Zhou WL, Potapenko ES, Kim EJ & Antic SD(2011). Brief dopaminergic stimulations produce transientphysiological changes in prefrontal pyramidal neurons.Brain Res 1370, 1–15.

Neve KA, Seamans JK & Trantham-Davidson H (2004).Dopamine receptor signalling. J Recept Signal Transduct Res24, 165–205.

Palmer LM & Stuart GJ (2006). Site of action potentialinitiation in layer 5 pyramidal neurons. J Neurosci 26,1854–1863.

Peterson JD, Wolf ME & White FJ (2006). Repeatedamphetamine administration decreases D1 dopaminereceptor-mediated inhibition of voltage-gated sodiumcurrents in the prefrontal cortex. J Neurosci 26, 3164–3168.

Popovic MA, Foust AJ, McCormick DA & Zecevic D (2011).The spatio-temporal characteristics of action potentialinitiation in layer 5 pyramidal neurons: a voltage imagingstudy. J Physiol 589, 4167–4187.

Sabatini BL, Maravall M & Svoboda K (2001). Ca2+ signallingin dendritic spines. Curr Opin Neurobiol 11, 349–356.

Sapolsky RM (2001). Cellular defenses against excitotoxicinsults. J Neurochem 76, 1601–1611.

Schultz W (2002). Getting formal with dopamine and reward.Neuron 36, 241–263.



Seamans JK & Yang CR (2004). The principal features andmechanisms of dopamine modulation in the prefrontalcortex. Prog Neurobiol 74, 1–58.

Sesack SR & Bunney BS (1989). Pharmacologicalcharacterization of the receptor mediatingelectrophysiological responses to dopamine in the rat medialprefrontal cortex: a microiontophoretic study. J PharmacolExp Ther 248, 1323–1333.

Sesack SR, Deutch AY, Roth RH & Bunney BS (1989).Topographical organization of the efferent projections of themedial prefrontal cortex in the rat: an anterogradetract-tracing study with Phaseolus vulgaris leucoagglutinin. JComp Neurol 290, 213–242.

Somogyi P, Tamas G, Lujan R & Buhl EH (1998). Salientfeatures of synaptic organisation in the cerebral cortex. BrainRes Brain Res Rev 26, 113–135.

Stone JM, Morrison PD & Pilowsky LS (2007). Glutamate anddopamine dysregulation in schizophrenia – a synthesis andselective review. J Psychopharmacol 21, 440–452.

Surmeier DJ, Bargas J, Hemmings HC Jr, Nairn AC &Greengard P (1995). Modulation of calcium currents by a D1dopaminergic protein kinase/phosphatase cascade in ratneostriatal neurons. Neuron 14, 385–397.

Tsubokawa H & Ross WN (1996). IPSPs modulate spikebackpropagation and associated [Ca2+]i changes in thedendrites of hippocampal CA1 pyramidal neurons. JNeurophysiol 76, 2896–2906.

Vogt KE, Gerharz S, Graham J & Canepari M (2011).Combining membrane potential imaging with L-glutamateor GABA photorelease. PLoS One 6, e24911.

Westenbroek RE, Hell JW, Warner C, Dubel SJ, Snutch TP &Catterall WA (1992). Biochemical properties and subcellulardistribution of an N-type calcium channel α1 subunit.Neuron 9, 1099–1115.

Yang CR & Seamans JK (1996). Dopamine D1 receptor actionsin layers V-VI rat prefrontal cortex neurons in vitro:modulation of dendritic-somatic signal integration. JNeurosci 16, 1922–1935.

Young CE & Yang CR (2004). Dopamine D1/D5 receptormodulates state-dependent switching of soma-dendriticCa2+ potentials via differential protein kinase A and Cactivation in rat prefrontal cortical neurons. J Neurosci 24,8–23.

Zhou WL, Yan P, Wuskell JP, Loew LM & Antic SD (2007).Intracellular long wavelength voltage-sensitive dyes forstudying the dynamics of action potentials in axons and thindendrites. J Neurosci Methods 164, 225–239.

Author contributions

Design and execution of the experiments, analysis of the data,drafting the manuscript, comments on drafts and approvalof the final manuscript: both authors. The experiments wereconducted in S.D.A.’s laboratory at University of ConnecticutHealth Center, Farmington, CT, USA.

Acknowledgments

This work was supported by National Institutes of Health grant(MH063503), and a NARSAD Young Investigator Award 2009.We are grateful to Corey D. Acker for Matlab programming,and Glenn Belinsky, Anna R. Moore and Matthew T. Rich forcomments.


rapid dopaminergic and gabaergic modulation of calcium and voltage transients in dendrites of...

Documents